resume fails to light display on Macbook Pro Retina on 3.8-rc1

[Greg KH]

On Tue, Feb 26, 2013 at 09:35:14AM -0800, Greg KH wrote:
> On Mon, Feb 25, 2013 at 07:45:45PM -0800, Greg KH wrote:
> > On Mon, Feb 25, 2013 at 02:32:43PM -0800, Greg KH wrote:
> > > On Mon, Feb 25, 2013 at 04:06:02PM +1000, Dave Airlie wrote:
> > > > On Mon, Feb 25, 2013 at 3:52 PM, Greg KH <gregkh at linuxfoundation.org> wrote:
> > > > > Hi Ben,
> > > > >
> > > > > My Macbook Pro Retina fails to resume properly on 3.8.  I tracked this
> > > > > down to commit 6c5a04249d7afeea3e0ed971e7813f84e29a1706 (drm/nvd0/disp:
> > > > > move link training helpers into core as display methods)
> > > > >
> > > > > Anything I can try to help solve this?
> > > > >
> > > > > Note, I'm using the Intel driver as the main controller for this laptop,
> > > > > well, I think I am, my xorg log is attached.
> > > > 
> > > > No you are using the nvidia, the efi always boots nvidia enabled now.
> > > 
> > > Really?  When did that change?  I thought I wanted to be using the Intel
> > > chip to save battery life.
> > > 
> > > > btw I just tested my drm-next tree on mine and it resumed the display
> > > > fine, something oopsed a few seconds later that I haven't tracked down
> > > > 
> > > > git://git.freedesktop.org/~airlied/linux drm-next
> > > > 
> > > > I'll be sending it to Linus this evening or tomorrow morning, once I
> > > > fix my tree.
> > > 
> > > Ok, I'll test again when it hits Linus's tree, and if that works, it
> > > would be good to try to work out what patch fixes it to get them into
> > > the 3.8-stable series so that others don't run into the same problem.
> > 
> > I've tested Linus's tree now (I'm guessing it has all of your changes in
> > it), and it works!
> > 
> > I see a bunch of patches marked for the stable branch, so I'll try those
> > out and see if they fix the problem.  If not, I'll let you and Ben know.
> 
> I've applied the radeon patches tagged for -stable and tested that on
> 3.8.0, but that doesn't solve the resume problem.  Any ideas of anything
> else I can do to test this?  Doing a "backwards" git-bisect is a pain,
> but I guess I can do that to try to track down what patch fixed this, if
> that's the only idea people have...

Ok, after getting my brain warped by doing a reverse 'git bisect', I
tracked it down to commit ac8cc241a81941932da44993242e68c62e115ec7
(drm/nv50/devinit: reverse the logic for running encoder init scripts)
as the patch that solves the problem.

So, I tried applying that patch on 3.8.1-rc1, but ran into the efifb bug
that Dave fixed already, which took me a while, so I've queued that up
for 3.8.1.  Then I determined that it really also needs commit
f3ed1048715f2edc10c4dda6148b60e93f6282ed (drm/nouveau/bios: parse
external transmitter type if off-chip) and commit
8e992c8d9eebc2bd3246252ee5c0422dbbbce7ae (drm/nouveau/bios: store a
type/mask hash in parsed dcb data) in order to build properly.

With those three patches, plus the efifb fix, I can now properly boot,
and suspend and resume my macbook on 3.8.1-rc1.  Just like I can on
Linus's tree.

So, Ben, Dave, any objection to me including these 3 nouveau patches in
the 3.8.1-final release?

thanks,

greg k-h